After bloom bulbs are harvested they are peel , sorted and checked on flaw . skin and classify the bulbs can cause damage to the bulb , sham the quality of the Cartesian product . In improver , some bulbs are miss during the manual quality ascendance and terminate up in the incorrect grade . The Business Unit Greenhouse Horticulture & Flower Bulbs of Wageningen University & Research is investigate if vision engineering and thick learning can be used to find out the volume of bulbs and pick out medulla oblongata with defects .
After blossom bulbs are harvested , a few processing steps are done before the bulbs are pack . The first step is to remove clump and old root , call peeling . They are first cleared of the clump and then flake off . The bulbs then end up on a screen , where gravid electric light are discriminate from modest ones . However , errors are possible : for example , oval shape small incandescent lamp can sometimes end up in the incorrect stratum . As a result , a tulip grower ( the buyer of the bulb grower ) can get mint that are not uniform and therefore flower unevenly .
WUR look into whether the intensity and size of a bulb can be properly determined with a 3D camera . Several hundred electric-light bulb were choose for the 3D photographic camera measurements . and the result were compared with a manual book measurement by concentration . This demonstrate that the volume measurement with the 3D tv camera compared to the manual measurements was very good ( correlativity of 0.92 ) .
Deep learningThe 3D television camera was used in compounding with deep encyclopedism as well to investigate whether the quality could be determined based on television camera images . The neuronic web checks whether the efflorescence bulb has been pare by rights , is damaged or not and whether the skin quality is good . Bulbs with those defects are sold as lower class products . In addition , a damaged lightbulb is more sore to pollution with fusarium .
These mensuration showed that the loudness / size and select " defective " flower lightbulb was potential using a 3D tv camera . In 2021 , WUR will look into whether this is also potential with bombastic volumes : can the photographic camera and mysterious learning still keep up ?
The inquiry took seat in the setting of Bulb Revolution 4.0 . The execution of this Public - Private Partnership lies with a pool of seven mate . These are : KAVB , Anthos , Wageningen University & Research ( WUR ) , Cremer Speciaalmachines B.V. , Machinefabriek Steketee B.V. , Agrisim B.V. , BKD and TechNature B.V. Economic Board Greenport Duin & Bollenstreek and Rabobank Bollenstreek contribute to the funding from their foundation Funds.
For more selective information : Wageningen University & Researchwww.wur.nl
